My hiring operating system, as code. A self-calibrating talent evaluation MCP server that encodes how I think about candidates. 6 dimensions, dynamic weights, and continuous learning.
Hiring decisions are the highest-leverage choices a company makes, yet most evaluation happens in people's heads, inconsistently, uncalibrated, and with no audit trail. Interviewers score candidates differently depending on their mood, recency bias, or how the last three interviews went. There's no system-level feedback loop.
RishiOS MCP is my hiring operating system, encoded as software. It plugs into AI-powered workflows (Claude Desktop, custom agents) and provides structured evaluation tools that run alongside me. Not replacing judgment, but keeping it calibrated.
Every candidate is evaluated on the same six dimensions: Systems Thinking, Agency, Craft Depth, Judgment, Communication, and Influence. What changes is the weight each dimension carries depending on the role and seniority level. A Staff Engineer gets heavier weighting on Systems Thinking; a Product Manager gets more on Influence.
Most TA leaders talk about "raising the bar," but without a system that actually tracks where the bar is, it's just aspiration. This tool gives me a real-time view of my own evaluation patterns, catches drift before it compounds, and makes every hiring decision more defensible.